An intelligent diagnosis method, product and device for the structural safety of the integrity of oil and gas pipelines

Abstract

The present invention provides an intelligent diagnosis method, product and device for the structural safety of the integrity of oil and gas pipelines. The intelligent diagnosis method for the structural safety of the integrity of oil and gas pipelines includes: obtaining more than one set of oil and gas pipeline characteristic data; inputting more than one set of oil and gas pipeline characteristic data into a pre-trained diagnosis model, and the diagnosis model outputs the integrity prediction level of each set of oil and gas pipeline characteristic data; the diagnosis model includes M weak regression learners corresponding one by one to M intermediate regression results, and a classification model; inputting each set of oil and gas pipeline characteristic data into the M weak regression learners to obtain the regression values of the M intermediate regression results corresponding to this set of oil and gas pipeline characteristic data, and the classification model obtains the integrity prediction level of each set of oil and gas pipeline characteristic data. The diagnosis model adopts a multi-level boosting model structure, which improves the accuracy of the integrity diagnosis of oil and gas pipelines and enhances the generalization ability and robustness of the diagnosis model in processing complex and variable data.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent monitoring and maintenance of oil and gas pipelines, and particularly to an intelligent diagnosis method, product and device for the structural safety of the integrity of oil and gas pipelines. Background Art

[0002] As a crucial link in the oil and gas industry chain, the safe operation of oil and gas pipelines is directly related to the stability of energy supply and environmental safety. During operation, oil and gas pipelines may be damaged due to factors such as material aging, external interference, and environmental corrosion, which may affect the integrity and operation safety of the pipelines. Therefore, timely and accurate diagnosis of the integrity status of oil and gas pipelines is of great significance for preventing accidents and ensuring the safe supply of energy.

[0003] Traditional pipeline integrity assessment methods mostly rely on regular physical inspections and manual patrols. These methods are not only time-consuming and laborious but also difficult to achieve real-time monitoring of the overall pipeline status. With the improvement of sensor technology, data collection capabilities, and the development of computing technology, data-driven oil and gas pipeline integrity diagnosis methods have begun to receive attention. These methods analyze a large amount of collected data and use machine learning algorithms to predict key parameters such as the corrosion rate and potential defects of the pipeline, thereby realizing real-time and dynamic assessment of pipeline integrity.

[0004] Although data-driven methods provide an effective means for pipeline integrity diagnosis, the existing technologies still face many challenges when dealing with complex and non-linear oil and gas pipeline data. The complexity of oil and gas pipeline data is mainly manifested in the multiple influencing factors contained in the data, such as the characteristics of the internal medium, pipeline material, and external environmental conditions. The interaction of these factors makes the data have high-dimensional and non-linear characteristics. In addition, the limitations of monitoring equipment and environmental interference may lead to missing and noisy data, which poses higher requirements for data preprocessing and feature extraction.

[0005] When applying traditional analysis techniques, such as linear regression models, principal component analysis (PCA), linear discriminant analysis (LDA), and basic statistical methods, for pipeline integrity diagnosis, although these methods can reveal the basic trends and patterns of the data to a certain extent, they perform poorly in terms of prediction accuracy and reliability. In addition, due to problems such as low computational efficiency or easy overfitting, the application of the above methods in oil and gas pipeline data analysis and prediction is limited and cannot fully exert their potential diagnostic capabilities. Summary of the Invention

[0006] The present invention aims to solve the technical problems in the existing oil and gas pipeline integrity diagnosis technology, such as poor prediction accuracy and weak generalization ability caused by easy overfitting, and provides an intelligent diagnosis method, product and device for the structural safety of the integrity of oil and gas pipelines.

[0007] To achieve the above object of the present invention, according to the first aspect of the present invention, the present invention provides an intelligent diagnosis method for the structural safety of the integrity of oil and gas pipelines, including: obtaining more than one set of oil and gas pipeline characteristic data, and each set of oil and gas pipeline characteristic data includes at least one of H 2 S characteristics, oil pressure characteristics, PH characteristics, production characteristics, and temperature characteristics; inputting the above-mentioned more than one set of oil and gas pipeline characteristic data into a pre-trained diagnosis model, and the diagnosis model outputs the integrity prediction level of the oil and gas pipeline for each set of oil and gas pipeline characteristic data; the diagnosis model includes M weak regression learners corresponding one by one to M intermediate regression results, and a classification model; inputting each set of oil and gas pipeline characteristic data into the M weak regression learners to obtain the regression values of the M intermediate regression results corresponding to this set of oil and gas pipeline characteristic data, and the classification model processes the regression values of the M intermediate regression results corresponding to each set of oil and gas pipeline characteristic data to obtain the integrity prediction level of the oil and gas pipeline for each set of oil and gas pipeline characteristic data; wherein, the M intermediate regression results include at least one of the corrosion rate regression result, the erosion rate regression result, the stiffness regression result, and the life regression result, and M is a positive integer.

[0008] To achieve the above object of the present invention, according to the second aspect of the present invention, the present invention provides a computer program product, including a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of the intelligent diagnosis method for the structural safety of the integrity of oil and gas pipelines described in the first aspect of the present invention are implemented.

[0009] To achieve the above object of the present invention, according to the third aspect of the present invention, the present invention provides an electronic device, and the electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores a computer program executable by the at least one processor, and when the computer program is executed by the at least one processor, the at least one processor can execute the intelligent diagnosis method for the structural safety of the integrity of oil and gas pipelines described in the first aspect of the present invention.

[0010] Advantageous technical effects of the present invention: The diagnostic model adopts a multi-level boosting model structure. The first level consists of M weak regression learners for multi-attribute regression, and the second level is a classification model for multi-attribute classification. The M weak regression learners in the first level respectively map each set of oil and gas pipeline feature data to the regression values of the corresponding intermediate regression results. The second level performs classification processing based on the regression values of the M intermediate regression results of this set of oil and gas pipeline feature data, fully capturing the correlation and dependence between the intermediate regression results, and obtaining a relatively accurate prediction level of the integrity of the oil and gas pipeline for this set of oil and gas pipeline feature data. The diagnostic model is trained through a boosting algorithm, which can effectively prevent the problem of overfitting in data training, improve the accuracy of oil and gas pipeline integrity diagnosis, can handle both classification and regression problems simultaneously, and enhances the generalization ability and robustness of the diagnostic model in dealing with complex and variable data by combining multiple weak regression learners. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 is a schematic flow chart of an intelligent diagnosis method for the structural safety of oil and gas pipeline integrity in a preferred embodiment of the present invention;

[0012] Figure 2 is a schematic structural diagram of the trained diagnostic model in a preferred embodiment of the present invention;

[0013] Figure 3 is a schematic diagram of the data processing process of the diagnostic model in a preferred embodiment of the present invention;

[0014] Figure 4 is a schematic diagram of the multi-attribute prediction and multi-attribute classification levels of the diagnostic model in a preferred embodiment of the present invention;

[0015] Figure 5 is a schematic structural diagram of the classification model in a preferred embodiment of the present invention;

[0016] Figure 6 is a schematic structural diagram of the electronic device in a preferred embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary and are only used to explain the present invention and should not be construed as limiting the present invention.

[0018] In the description of the present invention, it should be understood that the orientation or positional relationships indicated by the terms "longitudinal", "transverse", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. are based on the orientation or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention.

[0019] In the description of the present invention, unless otherwise specified and defined, it should be noted that the terms "installation", "connection", and "coupling" should be understood in a broad sense. For example, it can be a mechanical connection or an electrical connection, or it can be the communication inside two elements. It can be directly connected or indirectly connected through an intermediate medium. For those of ordinary skill in the art, the specific meanings of the above terms can be understood according to specific circumstances.

[0020] The present invention discloses an intelligent diagnostic method for the integrity structure safety of oil and gas pipelines. In a preferred embodiment, referring to Figure 1 , it includes:

[0021] Step S1, obtaining more than one set of oil and gas pipeline characteristic data, and each set of oil and gas pipeline characteristic data includes at least one of H 2 S characteristics, oil pressure characteristics, PH characteristics, production characteristics, and temperature characteristics.

[0022] In this embodiment, in order to better reflect the state of the oil and gas pipeline and obtain a more accurate prediction result of pipeline integrity, preferably, the oil and gas pipeline characteristic data includes static characteristic data and production dynamic characteristic data. The static characteristic data includes pipeline material characteristics, pipeline construction characteristics (such as pipeline installation slope, number of pipeline bends, etc.) and pipeline detection characteristics (such as pipeline strength). The production dynamic characteristic data includes, but is not limited to, temperature characteristics, oil pressure characteristics, set gas content, production characteristics, and PH characteristics in the pipeline, etc. The production specifically refers to the oil and gas transportation volume of the oil and gas pipeline within a preset time period. The preset time period is preferably but not limited to one day or two days, and can be customarily set.

[0023] Step S2, inputting more than one set of oil and gas pipeline characteristic data into a pre-trained diagnostic model, and the diagnostic model outputs the oil and gas pipeline integrity prediction level corresponding to each set of oil and gas pipeline characteristic data. The integrity prediction can be simultaneously based on multiple sets of oil and gas pipeline characteristic data to obtain their respective oil and gas pipeline integrity prediction levels. Exemplarily, the higher the oil and gas pipeline integrity prediction level, the more complete the oil and gas pipeline is, and the lower the oil and gas pipeline integrity prediction level, the more serious the damage, corrosion, etc. of the oil and gas pipeline are.

[0024] In this embodiment, referring to Figure 2, the diagnostic model includes M weak regression learners corresponding one-to-one to M intermediate regression results, and a classification model; inputting each group of oil and gas pipeline feature data into the M weak regression learners to obtain the regression values of the M intermediate regression results corresponding to this group of oil and gas pipeline feature data, and the classification model processes the regression values of the M intermediate regression results corresponding to each group of oil and gas pipeline feature data to obtain the oil and gas pipeline integrity prediction level of each group of oil and gas pipeline feature data; wherein, the M intermediate regression results include at least one of the corrosion rate regression result, erosion rate regression result, strength regression result, and life regression result, and M is a positive integer. Figure 4 shows the schematic diagram of the multi-level structure of the diagnostic model. The first layer is the multi-attribute regression stage, and the second layer is the multi-attribute classification stage.

[0025] In this embodiment, the weak regression learner is preferably but not limited to a multiple linear regressor, which has a weight vector obtained through iterative training. Each weak regression learner corresponds to an intermediate regression result. The classification model is preferably but not limited to an existing multi-classification model, such as an existing SVM classification model or Softmax multi-classification model. For a group of oil and gas pipeline feature data, each weak regression learner maps it to the regression value of its corresponding intermediate regression result, and the classification model performs classification processing based on the regression values of the M intermediate regression results to obtain the oil and gas pipeline integrity prediction level of this group of oil and gas pipeline feature data.

[0026] Figure 3 shows the data processing process of the diagnostic model for a group of oil and gas pipeline feature data. A group of oil and gas pipeline feature data is represented as {x 1 , x 2 ,..., x k ,..., x K}, including K features, and 1 ≤ k ≤ K, where K is a positive integer. The K features are respectively input into the M weak regression learners, and each weak regression learner outputs the regression value of the corresponding intermediate regression result {y 1 , y 2 ,..., y m ,..., y M}; 1 ≤ m ≤ M, where M is a positive integer. The regression values of the M intermediate regression results are input into the classification model together, and the classification model outputs the oil and gas pipeline integrity prediction level z of this group of oil and gas pipeline feature data.

[0027] In a preferred embodiment, since there are a large number of outliers and missing values in the original oil and gas pipeline data, it is necessary to eliminate noise and handle data missing anomalies. To improve the accuracy of oil and gas pipeline integrity prediction, the process of obtaining each group of oil and gas pipeline feature data is as follows:

[0028] Step S11, obtain a set of multi-dimensional original data of the oil and gas pipeline. The multi-dimensional original data includes static data and production dynamic data. The static data includes material data, construction data, and inspection data of the oil and gas pipeline. The production dynamic data includes temperature data, oil pressure data, casing pressure data, production volume, water production volume, hydrogen sulfide content, carbon dioxide content, and chemical composition data, etc. in the oil and gas pipeline.

[0029] Step S12, extract the original data corresponding to the preset dimension from the set of multi-dimensional original data, and record it as the preset dimension data.

[0030] Among the multiple dimensions of the original data, there is a situation where some dimensions contribute more to the integrity prediction of the oil and gas pipeline, and some dimensions contribute less to the integrity prediction of the oil and gas pipeline. Based on this, the preset dimension can be determined manually or through a preset analysis algorithm. The preset dimension preferably but not limited to includes the H 2 S content feature, oil pressure feature, PH feature, production volume feature, temperature in the pipeline.

[0031] Step S13, perform normalization processing on the preset dimension data to obtain normalized data.

[0032] Through normalization processing, the data scale can be unified. The normalization processing preferably but not limited to is converted by the maximum-minimum normalization method, so that all values are between 0 and 1, so that each feature value is in the same order of magnitude during the subsequent processing, and the deviation during the training of the diagnostic model is reduced.

[0033] Step S14, perform smoothing processing on the normalized data to obtain smoothed data.

[0034] Through smoothing processing, the random fluctuations in the data can be reduced. The smoothing processing preferably but not limited to performs smoothing processing on the normalized data through the exponential moving average method, which can effectively reduce the random fluctuations in the data, make the data trend more obvious, and facilitate subsequent identification and analysis.

[0035] Step S15, perform missing value supplementation and / or outlier deletion on the smoothed data to obtain a set of oil and gas pipeline feature data.

[0036] Interpolate and supplement the missing pipeline data, and delete the outliers in the pipeline data, so as to reduce the interference during the training of the diagnostic model.

[0037] In this embodiment, in order to more accurately determine the preset dimension, further preferably, the preset analysis algorithm for the preset dimension includes:

[0038] Step S01, obtain the multi-dimensional original data of the oil and gas pipeline. The multi-dimensional original data can be composed of one set or multiple sets of multi-dimensional original data.

[0039] Step S02: Calculate the correlation coefficients between the original data of each dimension and the predicted integrity level of the oil and gas pipeline respectively, and select the dimensions with correlation coefficients greater than or equal to the preset correlation threshold as the preset dimensions.

[0040] In this embodiment, preferably but not limited to, the existing Person correlation analysis algorithm is used to analyze the correlation coefficients between the data of each dimension and the predicted integrity level of the oil and gas pipeline. The preset correlation threshold is 0.8, and the dimensions corresponding to the correlation coefficients in the range of 0.8 - 1 are selected as a preset dimension, and the data corresponding to this preset dimension is used as a part of the characteristic data of the oil and gas pipeline.

[0041] In one example, the original data includes 20 - dimensional information, where the H 2 S content, oil pressure, pH value, and temperature inside the pipeline all have correlation coefficients greater than the preset correlation threshold with the predicted integrity level of the oil and gas pipeline. The H 2 S content, oil pressure, pH value, production, and temperature are used as preset dimensions. Stripping the above five - dimensional data for diagnostic model training will reduce the interference to the diagnostic model and improve the prediction accuracy of the diagnostic model.

[0042] In a preferred embodiment, for the traditional gradient - boosting algorithm, since gradient boosting needs to sequentially add weak learners, when facing large - scale data or a large number of trees, the training time will increase significantly, and overfitting and other problems are likely to occur. This application proposes Figure 4 a diagnostic model with a multi - level structure as shown, which is improved from two aspects: parallel execution of multiple weak regression learners and phased completion of regression prediction and classification tasks. Specifically, the training method of the diagnostic model is as follows:

[0043] Step 1: Obtain multiple groups of multi - dimensional original data of the oil and gas pipeline, convert the multiple groups of multi - dimensional original data into a sample set, and divide the sample set into a regression training set and a regression validation set. In one example, multiple groups of multi - dimensional original data of the oil and gas pipeline collected by 20 sensors in 4 days are selected. Each sensor collects data every 10 seconds, and a total of 691,200 groups of data are collected.

[0044] In Step 1, for each group of multi - dimensional original data, execute Steps S11 - S15 to obtain a corresponding group of oil and gas pipeline characteristic data, and use it as a sample. Set a sample label for each sample. The sample label includes the true values of the M intermediate regression results corresponding to the sample, as well as the true integrity level of the oil and gas pipeline. Preferably but not limited to, divide the sample set into a regression training set and a regression validation set according to a ratio of 7:3.

[0045] Step 2: Iteratively train M weak regression learners using the regression training set, and verify the M weak regression learners after the iterative training using the regression validation set. Continuously update the weight vectors of the weak regression learners during the iterative training.

[0046] Step 3: Input the regression training set into the M weak regression learners that have passed the verification again to obtain the classification training set, and input the regression validation set into the M weak regression learners that have passed the verification to obtain the classification validation set.

[0047] Input the regression training set into the M weak regression learners that have passed the verification again. Each regression training sample obtains the regression values of M intermediate regression results, and the M regression values form a classification training sample. Input the regression validation set into the M weak regression learners that have passed the verification again. Each regression validation sample obtains the regression values of M intermediate regression results, and the M regression values form a classification validation sample.

[0048] Step 4: Train the classification model using the classification training set, and verify the trained classification model using the classification validation set. The M weak regression learners that have passed the verification obtained in Step 2 and the classification model that has passed the verification together form the trained diagnostic model.

[0049] In this embodiment, preferably, iteratively training M weak regression learners using the regression training set includes:

[0050] Initialize the weight vectors of the M weak regression learners respectively. The weight vector includes K weight values, and the K weight values respectively correspond to K input feature data dimensions.

[0051] In each iterative training, input all samples in the regression training set into each weak regression learner at the same time, and the M weak regression learners perform iterative training. Preferably, the M weak regression learners perform iterative training in parallel. Among them, the t-th iterative training process of the m-th weak regression learner includes:

[0052] Step A1: Input the regression training set into the m-th weak regression learner, and the m-th weak regression learner outputs the regression values of the m-th intermediate regression results of each regression training sample.

[0053] Step A2: Calculate the regression loss of the m-th weak regression learner.

[0054] Step A3: If the regression loss reaches the preset threshold, that is, the regression loss is less than or equal to the preset threshold, stop the iterative training of the m-th weak regression learner (that is, no longer train the m-th weak regression learner in the (t + 1)-th iterative training), and save the weight vector of the t-th iterative training of the m-th weak regression learner. If the regression loss does not reach the preset threshold, that is, the regression loss is greater than the preset threshold, execute Step A4.

[0055] Step A4, calculate the gradient of the regression loss function with respect to the regression value of the m-th weak regression learner, and obtain the weight vector of the (t + 1)-th iterative training of the m-th weak regression learner according to this gradient by the weight update formula;

[0056] where m is a positive integer and 1 ≤ m ≤ M.

[0057] In this embodiment, the weight update formula is:

[0058]

[0059] where ω (t+1) is the weight vector of the (t + 1)-th round of iterative training of the weak regression learner; is the weight vector of the t-th iteration of the weak regression learner; L(y i , f(x i )) is the regression loss function, preferably but not limited to representing the true value y i of the m-th intermediate regression result of the i-th regression training sample randomly selected from the t-th iterative training and the difference between the regression value f(x i ) of the m-th intermediate regression result of the i-th regression training sample; is the gradient of the regression loss function with respect to f(x i ).

[0060] In this embodiment, a series of weak regression learners are iteratively trained, and each round focuses on correcting the errors left by the previous round.

[0061] In this embodiment, the multi-attribute regression algorithm corresponding to layer one inherits the boosting algorithm. Combine multiple weak learners whose performance is only slightly better than random guessing to form a strong learner with high precision.

[0062] In this embodiment, further preferably, the regression loss of the m-th weak regression learner is:

[0063]

[0064] where L o ss m，t represents the regression loss of the m-th weak regression learner in the t-th iterative training; N represents the total number of samples in the regression training set; i represents the regression training sample index; y m，t，i represents the regression value of the m-th intermediate regression result corresponding to the i-th sample output by the m-th weak regression learner in the t-th iterative training; represents the true value corresponding to the m-th intermediate regression result of the i-th sample; ω m，t represents the weight vector of the t-th iterative training of the m-th weak regression learner; ||ωm，t || represents the modulus of ω m，t ; λ is the regularization parameter used to control the strength of regularization; δ represents the prediction deviation threshold; f σ，i is the first function.

[0065] In this embodiment, in the above regression loss function, the LOG logarithmic function is adopted, which can smooth the output of the regression loss function and make the regression loss function more stable when dealing with large errors. cosh is the hyperbolic cosine function, which is an even function and non - negative for all real numbers. Summing the squares of the weight vectors of all weak regression learners, ||ω j || is the modulus of the weight vector of the current weak regression learner. f σ，i is the first function, when it is the case, the value of this function is 0, when it is the case, the value of this function is 1, so that the regularization term can be dynamically selected according to the magnitude of the error between the predicted value and the true value. In the above improved Log - Cosh regression loss function, the hyperbolic cosine function helps to smooth the prediction error. For small errors, its behavior is similar to the mean square error, while for large errors, its growth rate is slower than the square error, which makes the improved Log - Cosh regression loss more robust to outliers. The regularization term can effectively control the complexity of the weak regression learner. By penalizing large weight vectors, it prompts the weak regression learner to prefer smaller and more dispersed weight vectors, which helps to improve the prediction accuracy of the weak regression learner for unseen data and prevents the weak regression learner from relying too much on a few features in the training data, thus reducing the risk of overfitting. In summary, the improved gradient boosting regression algorithm enhances the robustness and smoothness and reduces the risk of gradient explosion.

[0066] In a preferred embodiment, to capture the correlation and dependence between different intermediate regression results to ensure better classification by the classification model, thereby improving the prediction performance of the diagnostic model, an iterative feature extraction module is designed to capture the correlation and dependence between intermediate regression results. Referring to Figure 5 , the classification model includes an iterative feature extraction module and an SVM classifier; the iterative feature extraction module performs: inputting the regression values of the M intermediate regression results of each group of oil and gas pipeline feature data into the dynamic focusing network to obtain the extracted features, repeating Q times inputting the extracted features output by the dynamic focusing network back into the dynamic focusing network to obtain new extracted features, and outputting the extracted features output by the dynamic focusing network after the Q - th repetition to the SVM classifier, where Q is an integer greater than or equal to 0; the SVM classifier outputs the oil and gas pipeline integrity prediction level of each group of oil and gas pipeline feature data. Q is preferably but not limited to 4.

[0067] In this embodiment, further preferably, referring toFigure 5 , the dynamic focusing network includes a multi-layer perceptron, a pooling layer, a first convolutional layer, a second convolutional layer, and an activation output unit that are connected in sequence.

[0068] In this embodiment, preferably, the multi-layer perceptron includes three hidden layers. The pooling layer, the first convolutional layer, the second convolutional layer, and the activation output unit form a dynamic focusing unit. Through the multi-layer perceptron and the dynamic focusing unit, the non-linear processing ability and stability of the diagnostic model are enhanced.

[0069] In this embodiment, the regression values of the input M intermediate regression results form a regression value sequence or the extracted features obtained in the previous iteration. First, they are subjected to preliminary feature extraction by the multi-layer perceptron. The features extracted by the multi-layer perceptron are sent to the dynamic focusing unit. The dynamic focusing unit is designed with a pooling layer, a first convolutional layer, and a second convolutional layer. The activation output unit uses the relu activation function to further process and refine the features. In addition, the dynamic focusing unit dynamically weights the features, highlighting important features and suppressing secondary features, so as to more accurately capture the information crucial for the classification decision of the SVM classifier. The dynamic focusing module greatly improves the performance and efficiency of the model in processing various types of data, especially tabular data, by assigning different weights to different parts of the input data, enabling the SVM classifier to focus on processing more important information regions, thereby increasing the attention and the sensitivity of the SVM classifier to key information. In tabular data, the SVM classifier can preferentially process the features that have the greatest impact on the result, significantly improving the accuracy of the SVM classifier's decision and the interpretability of the SVM classifier. The dynamic focusing unit can capture the dependence relationship between the M intermediate regression results, which helps to improve the prediction performance of the diagnostic model.

[0070] The present invention also discloses a computer program product, including a computer program / instructions. When the computer program / instructions are executed by a processor, the steps of the above-mentioned intelligent diagnosis method for the integrity structural safety of oil and gas pipelines provided by the present invention are implemented. The computer program product should be understood as a software product that mainly implements its solution through a computer program, such as a program product integrated in the cloud or a software library or a terminal device.

[0071] The present invention also discloses an electronic device. In one embodiment, the electronic device includes at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the intelligent diagnosis method for the integrity structural safety of oil and gas pipelines provided by the present invention.

[0072] Such as Figure 6As shown, it is a schematic structural diagram of an electronic device for an intelligent diagnosis method of the structural safety of the integrity of an oil and gas pipeline provided by an embodiment of the present invention. The electronic device may include a processor 10, a memory 11, a communication bus 12, and a communication interface 13, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as an intelligent diagnosis method program for the structural safety of the integrity of an oil and gas pipeline.

[0073] Among them, the processor 10 may be composed of integrated circuits in some embodiments. For example, it may be composed of a single packaged integrated circuit, or may be composed of multiple packaged integrated circuits with the same or different functions, including a combination of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control core (Control Unit) of the electronic device, connecting various components of the entire electronic device through various interfaces and lines, and by running or executing programs or modules stored in the memory 11 (such as executing an intelligent diagnosis method for the structural safety of the integrity of an oil and gas pipeline, etc.), and calling data stored in the memory 11, to perform various functions of the electronic device and process data.

[0074] The memory 11 includes at least one type of readable storage medium. The readable storage medium includes flash memory, mobile hard disks, multimedia cards, card-type memories (such as SD or DX memories, etc.), magnetic memories, magnetic disks, optical discs, etc. The memory 11 may be an internal storage unit of the electronic device in some embodiments, such as the mobile hard disk of the electronic device. The memory 11 may also be an external storage device of the electronic device in other embodiments, such as a plug-in mobile hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device. Further, the memory 11 may include both an internal storage unit and an external storage device of the electronic device. The memory 11 can not only be used to store application software installed on the electronic device and various types of data, such as the code of an intelligent diagnosis method program for the structural safety of the integrity of an oil and gas pipeline, etc., but also be used to temporarily store data that has been output or will be output.

[0075] The communication bus 12 can be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable connection communication between the memory 11, at least one processor 10, and the like.

[0076] The communication interface 13 is used for communication between the above-mentioned electronic device and other devices, including a network interface and a user interface. Optionally, the network interface can include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between this electronic device and other electronic devices. The user interface can be a display, an input unit (such as a keyboard), and optionally, the user interface can also be a standard wired interface or a wireless interface. Optionally, in some embodiments, the display can be an LED display, a liquid crystal display, a touch liquid crystal display, and an OLED (Organic Light-Emitting Diode) toucher, etc. Among them, the display can also be appropriately referred to as a display screen or a display unit, which is used to display the information processed in the electronic device and to display a visual user interface.

[0077] Figure 6 Only the electronic device with components is shown. Those skilled in the art can understand that Figure 6 the shown structure does not constitute a limitation on the electronic device. It may include fewer or more components than shown, or combine certain components, or have a different component arrangement. For example, although not shown, the electronic device may further include a power source (such as a battery) for powering each component. Preferably, the power source can be logically connected to at least one processor 10 through a power management device, so as to implement functions such as charge management, discharge management, and power consumption management through the power management device. The power source may also include any components such as one or more DC or AC power sources, a recharge device, a power failure detection circuit, a power converter or an inverter, and a power status indicator. The electronic device may also include various sensors, a Bluetooth module, a Wi-Fi module, etc., which will not be elaborated here.

[0078] It should be understood that the embodiments are only for illustration purposes and are not limited by this structure in the scope of the patent application.

[0079] Furthermore, if the modules / units integrated in the electronic device 1 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disc, a computer memory, a read-only memory (ROM, Read-Only Memory).

[0080] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", "one implementation manner", "one preferred implementation manner" or "some examples", etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0081] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and purposes of the present invention, and the scope of the present invention is defined by the claims and their equivalents.

Claims

1. An intelligent diagnostic method for oil and gas pipeline integrity structure safety, characterized in that: include: Acquire more than one set of oil and gas pipeline characteristic data, each set of oil and gas pipeline characteristic data includes at least one of H2S characteristics, oil pressure characteristics, production characteristics, PH characteristics, and temperature characteristics; Inputting the one or more sets of oil and gas pipeline characteristic data into a pre-trained diagnostic model, wherein the diagnostic model outputs a predicted level of oil and gas pipeline integrity for each set of oil and gas pipeline characteristic data; The diagnostic model includes M weak regression learners corresponding to the M intermediate regression results one by one, and a classification model; each group of oil and gas pipeline characteristic data is input into the M weak regression learners to obtain the regression values ​​of the M intermediate regression results corresponding to the group of oil and gas pipeline characteristic data; the classification model processes the regression values ​​of the M intermediate regression results corresponding to each group of oil and gas pipeline characteristic data to obtain the oil and gas pipeline integrity prediction level of each group of oil and gas pipeline characteristic data; The M intermediate regression results include at least one of the corrosion rate regression result, the erosion rate regression result, the rigidity regression result and the life regression result, and M is a positive integer; Among them, the regression training set is used to iteratively train M weak regression learners, including: Initialize the weight vectors of M weak regression learners respectively; M weak regression learners are iteratively trained, wherein the t-th iterative training process of the m-th weak regression learner includes: Step A1, inputting the regression training set into the mth weak regression learner, and the mth weak regression learner outputs the regression value of the mth intermediate regression result of each regression training sample; Step A2, calculate the regression loss of the mth weak regression learner as: Among them, Loss m,t represents the regression loss of the mth weak regression learner in the tth iteration training; N represents the total number of samples in the regression training set; i represents the regression training sample index; y m,t,i It represents the regression value of the mth intermediate regression result corresponding to the i-th sample output by the mth weak regression learner in the t-th iteration training; represents the true value corresponding to the mth intermediate regression result of the ith sample; ω m,t represents the weight vector of the t-th iteration training of the m-th weak regression learner; ||ω m,t || represents ω m,t The modulus of ; λ is the regularization parameter; δ represents the prediction deviation threshold; f σ,i is the first function; Step A3: if the regression loss reaches a preset threshold, stop the iterative training of the mth weak regression learner, and save the weight vector of the tth iterative training of the mth weak regression learner; if the regression loss does not reach the preset threshold, execute step A4; Step A4, calculating the gradient of the regression loss function with respect to the regression value of the m-th weak regression learner, and obtaining the weight vector of the t+1-th iteration training of the m-th weak regression learner according to the gradient and the weight update formula; Wherein, m is a positive integer, and 1≤m≤M.

2. The intelligent diagnosis method for oil and gas pipeline integrity structure safety according to claim 1, characterized in that: The process of obtaining each set of oil and gas pipeline characteristic data is as follows: Obtain a set of multi-dimensional raw data of oil and gas pipelines; Extracting original data corresponding to a preset dimension from a set of multi-dimensional original data, and recording the data as preset dimension data; Normalizing the preset dimension data to obtain normalized data; Smoothing the normalized data to obtain smoothed data; The missing values ​​are supplemented and / or outliers are deleted for the smoothed data to obtain a set of oil and gas pipeline characteristic data.

3. The intelligent diagnosis method for oil and gas pipeline integrity structure safety according to claim 2, characterized in that: The method for determining the preset dimension is: Obtain multi-dimensional raw data of oil and gas pipelines; The correlation coefficient between the original data of each dimension and the predicted level of oil and gas pipeline integrity is calculated respectively, and the dimension with a correlation coefficient greater than or equal to a preset correlation threshold is selected as the preset dimension.

4. The intelligent diagnosis method for oil and gas pipeline integrity structure safety according to any one of claims 1 to 3, characterized in that: The training method of the diagnostic model is: Acquire multiple groups of multi-dimensional raw data of oil and gas pipelines, convert the multiple groups of multi-dimensional raw data into sample sets, and divide the sample sets into regression training sets and regression verification sets; Iteratively train M weak regression learners using the regression training set, and verify the M weak regression learners after the iterative training using the regression verification set; The regression training set is input again into the M weak regression learners that have passed the verification to obtain the classification training set, and the regression verification set is input into the M weak regression learners that have passed the verification to obtain the classification verification set; The classification training set is used to train the classification model, and the classification validation set is used to validate the trained classification model.

5. The intelligent diagnosis method for oil and gas pipeline integrity structure safety according to claim 4, characterized in that: The classification model includes an iterative feature extraction module and a SVM classifier; The iterative feature extraction module performs: inputting the regression values ​​of the M intermediate regression results of each group of oil and gas pipeline feature data into the dynamic focusing network to obtain extracted features, repeating Q times to re-input the extracted features output by the dynamic focusing network into the dynamic focusing network to obtain new extracted features, and outputting the extracted features output by the dynamic focusing network after the Qth repetition to the SVM classifier, where Q is an integer greater than or equal to 0; The SVM classifier outputs the predicted level of oil and gas pipeline integrity for each set of oil and gas pipeline feature data.

6. The intelligent diagnosis method for oil and gas pipeline integrity structure safety according to claim 5, characterized in that: The dynamic focusing network includes a multi-layer perceptron, a pooling layer, a first convolutional layer, a second convolutional layer and an activation output unit which are connected in sequence.

7. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

8. An electronic device, characterized in that: The electronic device comprises: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the intelligent diagnosis method for structural safety of oil and gas pipeline integrity as described in any one of claims 1 to 6.

Images